1 //===- Wasm.h - Wasm object file implementation -----------------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // This file declares the WasmObjectFile class, which implements the ObjectFile
10 // interface for Wasm files.
11 //
12 // See: https://github.com/WebAssembly/design/blob/master/BinaryEncoding.md
13 //
14 //===----------------------------------------------------------------------===//
15 
16 #ifndef LLVM_OBJECT_WASM_H
17 #define LLVM_OBJECT_WASM_H
18 
19 #include "llvm/ADT/ArrayRef.h"
20 #include "llvm/ADT/StringRef.h"
21 #include "llvm/BinaryFormat/Wasm.h"
22 #include "llvm/Config/llvm-config.h"
23 #include "llvm/MC/MCSymbolWasm.h"
24 #include "llvm/Object/Binary.h"
25 #include "llvm/Object/ObjectFile.h"
26 #include "llvm/Support/Error.h"
27 #include "llvm/Support/MemoryBuffer.h"
28 #include <cstddef>
29 #include <cstdint>
30 #include <vector>
31 
32 namespace llvm {
33 namespace object {
34 
35 class WasmSymbol {
36 public:
WasmSymbol(const wasm::WasmSymbolInfo & Info,const wasm::WasmGlobalType * GlobalType,const wasm::WasmTableType * TableType,const wasm::WasmEventType * EventType,const wasm::WasmSignature * Signature)37   WasmSymbol(const wasm::WasmSymbolInfo &Info,
38              const wasm::WasmGlobalType *GlobalType,
39              const wasm::WasmTableType *TableType,
40              const wasm::WasmEventType *EventType,
41              const wasm::WasmSignature *Signature)
42       : Info(Info), GlobalType(GlobalType), TableType(TableType),
43         EventType(EventType), Signature(Signature) {}
44 
45   const wasm::WasmSymbolInfo &Info;
46   const wasm::WasmGlobalType *GlobalType;
47   const wasm::WasmTableType *TableType;
48   const wasm::WasmEventType *EventType;
49   const wasm::WasmSignature *Signature;
50 
isTypeFunction()51   bool isTypeFunction() const {
52     return Info.Kind == wasm::WASM_SYMBOL_TYPE_FUNCTION;
53   }
54 
isTypeTable()55   bool isTypeTable() const { return Info.Kind == wasm::WASM_SYMBOL_TYPE_TABLE; }
56 
isTypeData()57   bool isTypeData() const { return Info.Kind == wasm::WASM_SYMBOL_TYPE_DATA; }
58 
isTypeGlobal()59   bool isTypeGlobal() const {
60     return Info.Kind == wasm::WASM_SYMBOL_TYPE_GLOBAL;
61   }
62 
isTypeSection()63   bool isTypeSection() const {
64     return Info.Kind == wasm::WASM_SYMBOL_TYPE_SECTION;
65   }
66 
isTypeEvent()67   bool isTypeEvent() const { return Info.Kind == wasm::WASM_SYMBOL_TYPE_EVENT; }
68 
isDefined()69   bool isDefined() const { return !isUndefined(); }
70 
isUndefined()71   bool isUndefined() const {
72     return (Info.Flags & wasm::WASM_SYMBOL_UNDEFINED) != 0;
73   }
74 
isBindingWeak()75   bool isBindingWeak() const {
76     return getBinding() == wasm::WASM_SYMBOL_BINDING_WEAK;
77   }
78 
isBindingGlobal()79   bool isBindingGlobal() const {
80     return getBinding() == wasm::WASM_SYMBOL_BINDING_GLOBAL;
81   }
82 
isBindingLocal()83   bool isBindingLocal() const {
84     return getBinding() == wasm::WASM_SYMBOL_BINDING_LOCAL;
85   }
86 
getBinding()87   unsigned getBinding() const {
88     return Info.Flags & wasm::WASM_SYMBOL_BINDING_MASK;
89   }
90 
isHidden()91   bool isHidden() const {
92     return getVisibility() == wasm::WASM_SYMBOL_VISIBILITY_HIDDEN;
93   }
94 
getVisibility()95   unsigned getVisibility() const {
96     return Info.Flags & wasm::WASM_SYMBOL_VISIBILITY_MASK;
97   }
98 
99   void print(raw_ostream &Out) const;
100 
101 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
102   LLVM_DUMP_METHOD void dump() const;
103 #endif
104 };
105 
106 struct WasmSection {
107   WasmSection() = default;
108 
109   uint32_t Type = 0;         // Section type (See below)
110   uint32_t Offset = 0;       // Offset with in the file
111   StringRef Name;            // Section name (User-defined sections only)
112   uint32_t Comdat = UINT32_MAX; // From the "comdat info" section
113   ArrayRef<uint8_t> Content; // Section content
114   std::vector<wasm::WasmRelocation> Relocations; // Relocations for this section
115 };
116 
117 struct WasmSegment {
118   uint32_t SectionOffset;
119   wasm::WasmDataSegment Data;
120 };
121 
122 class WasmObjectFile : public ObjectFile {
123 
124 public:
125   WasmObjectFile(MemoryBufferRef Object, Error &Err);
126 
127   const wasm::WasmObjectHeader &getHeader() const;
128   const WasmSymbol &getWasmSymbol(const DataRefImpl &Symb) const;
129   const WasmSymbol &getWasmSymbol(const SymbolRef &Symbol) const;
130   const WasmSection &getWasmSection(const SectionRef &Section) const;
131   const wasm::WasmRelocation &getWasmRelocation(const RelocationRef &Ref) const;
132 
classof(const Binary * v)133   static bool classof(const Binary *v) { return v->isWasm(); }
134 
dylinkInfo()135   const wasm::WasmDylinkInfo &dylinkInfo() const { return DylinkInfo; }
getProducerInfo()136   const wasm::WasmProducerInfo &getProducerInfo() const { return ProducerInfo; }
getTargetFeatures()137   ArrayRef<wasm::WasmFeatureEntry> getTargetFeatures() const {
138     return TargetFeatures;
139   }
types()140   ArrayRef<wasm::WasmSignature> types() const { return Signatures; }
functionTypes()141   ArrayRef<uint32_t> functionTypes() const { return FunctionTypes; }
imports()142   ArrayRef<wasm::WasmImport> imports() const { return Imports; }
tables()143   ArrayRef<wasm::WasmTable> tables() const { return Tables; }
memories()144   ArrayRef<wasm::WasmLimits> memories() const { return Memories; }
globals()145   ArrayRef<wasm::WasmGlobal> globals() const { return Globals; }
events()146   ArrayRef<wasm::WasmEvent> events() const { return Events; }
exports()147   ArrayRef<wasm::WasmExport> exports() const { return Exports; }
syms()148   ArrayRef<WasmSymbol> syms() const { return Symbols; }
linkingData()149   const wasm::WasmLinkingData &linkingData() const { return LinkingData; }
getNumberOfSymbols()150   uint32_t getNumberOfSymbols() const { return Symbols.size(); }
elements()151   ArrayRef<wasm::WasmElemSegment> elements() const { return ElemSegments; }
dataSegments()152   ArrayRef<WasmSegment> dataSegments() const { return DataSegments; }
functions()153   ArrayRef<wasm::WasmFunction> functions() const { return Functions; }
debugNames()154   ArrayRef<wasm::WasmDebugName> debugNames() const { return DebugNames; }
startFunction()155   uint32_t startFunction() const { return StartFunction; }
getNumImportedGlobals()156   uint32_t getNumImportedGlobals() const { return NumImportedGlobals; }
getNumImportedTables()157   uint32_t getNumImportedTables() const { return NumImportedTables; }
getNumImportedFunctions()158   uint32_t getNumImportedFunctions() const { return NumImportedFunctions; }
getNumImportedEvents()159   uint32_t getNumImportedEvents() const { return NumImportedEvents; }
getNumSections()160   uint32_t getNumSections() const { return Sections.size(); }
161   void moveSymbolNext(DataRefImpl &Symb) const override;
162 
163   Expected<uint32_t> getSymbolFlags(DataRefImpl Symb) const override;
164 
165   basic_symbol_iterator symbol_begin() const override;
166 
167   basic_symbol_iterator symbol_end() const override;
168   Expected<StringRef> getSymbolName(DataRefImpl Symb) const override;
169 
170   Expected<uint64_t> getSymbolAddress(DataRefImpl Symb) const override;
171   uint64_t getWasmSymbolValue(const WasmSymbol &Sym) const;
172   uint64_t getSymbolValueImpl(DataRefImpl Symb) const override;
173   uint32_t getSymbolAlignment(DataRefImpl Symb) const override;
174   uint64_t getCommonSymbolSizeImpl(DataRefImpl Symb) const override;
175   Expected<SymbolRef::Type> getSymbolType(DataRefImpl Symb) const override;
176   Expected<section_iterator> getSymbolSection(DataRefImpl Symb) const override;
177   uint32_t getSymbolSectionId(SymbolRef Sym) const;
178 
179   // Overrides from SectionRef.
180   void moveSectionNext(DataRefImpl &Sec) const override;
181   Expected<StringRef> getSectionName(DataRefImpl Sec) const override;
182   uint64_t getSectionAddress(DataRefImpl Sec) const override;
183   uint64_t getSectionIndex(DataRefImpl Sec) const override;
184   uint64_t getSectionSize(DataRefImpl Sec) const override;
185   Expected<ArrayRef<uint8_t>>
186   getSectionContents(DataRefImpl Sec) const override;
187   uint64_t getSectionAlignment(DataRefImpl Sec) const override;
188   bool isSectionCompressed(DataRefImpl Sec) const override;
189   bool isSectionText(DataRefImpl Sec) const override;
190   bool isSectionData(DataRefImpl Sec) const override;
191   bool isSectionBSS(DataRefImpl Sec) const override;
192   bool isSectionVirtual(DataRefImpl Sec) const override;
193   relocation_iterator section_rel_begin(DataRefImpl Sec) const override;
194   relocation_iterator section_rel_end(DataRefImpl Sec) const override;
195 
196   // Overrides from RelocationRef.
197   void moveRelocationNext(DataRefImpl &Rel) const override;
198   uint64_t getRelocationOffset(DataRefImpl Rel) const override;
199   symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override;
200   uint64_t getRelocationType(DataRefImpl Rel) const override;
201   void getRelocationTypeName(DataRefImpl Rel,
202                              SmallVectorImpl<char> &Result) const override;
203 
204   section_iterator section_begin() const override;
205   section_iterator section_end() const override;
206   uint8_t getBytesInAddress() const override;
207   StringRef getFileFormatName() const override;
208   Triple::ArchType getArch() const override;
209   SubtargetFeatures getFeatures() const override;
210   bool isRelocatableObject() const override;
211   bool isSharedObject() const;
212 
213   struct ReadContext {
214     const uint8_t *Start;
215     const uint8_t *Ptr;
216     const uint8_t *End;
217   };
218 
219 private:
220   bool isValidFunctionIndex(uint32_t Index) const;
221   bool isDefinedFunctionIndex(uint32_t Index) const;
222   bool isValidGlobalIndex(uint32_t Index) const;
223   bool isValidTableIndex(uint32_t Index) const;
224   bool isDefinedGlobalIndex(uint32_t Index) const;
225   bool isDefinedTableIndex(uint32_t Index) const;
226   bool isValidEventIndex(uint32_t Index) const;
227   bool isDefinedEventIndex(uint32_t Index) const;
228   bool isValidFunctionSymbol(uint32_t Index) const;
229   bool isValidTableSymbol(uint32_t Index) const;
230   bool isValidGlobalSymbol(uint32_t Index) const;
231   bool isValidEventSymbol(uint32_t Index) const;
232   bool isValidDataSymbol(uint32_t Index) const;
233   bool isValidSectionSymbol(uint32_t Index) const;
234   wasm::WasmFunction &getDefinedFunction(uint32_t Index);
235   const wasm::WasmFunction &getDefinedFunction(uint32_t Index) const;
236   wasm::WasmGlobal &getDefinedGlobal(uint32_t Index);
237   wasm::WasmEvent &getDefinedEvent(uint32_t Index);
238 
239   const WasmSection &getWasmSection(DataRefImpl Ref) const;
240   const wasm::WasmRelocation &getWasmRelocation(DataRefImpl Ref) const;
241   uint32_t getSymbolSectionIdImpl(const WasmSymbol &Symb) const;
242 
243   Error parseSection(WasmSection &Sec);
244   Error parseCustomSection(WasmSection &Sec, ReadContext &Ctx);
245 
246   // Standard section types
247   Error parseTypeSection(ReadContext &Ctx);
248   Error parseImportSection(ReadContext &Ctx);
249   Error parseFunctionSection(ReadContext &Ctx);
250   Error parseTableSection(ReadContext &Ctx);
251   Error parseMemorySection(ReadContext &Ctx);
252   Error parseEventSection(ReadContext &Ctx);
253   Error parseGlobalSection(ReadContext &Ctx);
254   Error parseExportSection(ReadContext &Ctx);
255   Error parseStartSection(ReadContext &Ctx);
256   Error parseElemSection(ReadContext &Ctx);
257   Error parseCodeSection(ReadContext &Ctx);
258   Error parseDataSection(ReadContext &Ctx);
259   Error parseDataCountSection(ReadContext &Ctx);
260 
261   // Custom section types
262   Error parseDylinkSection(ReadContext &Ctx);
263   Error parseNameSection(ReadContext &Ctx);
264   Error parseLinkingSection(ReadContext &Ctx);
265   Error parseLinkingSectionSymtab(ReadContext &Ctx);
266   Error parseLinkingSectionComdat(ReadContext &Ctx);
267   Error parseProducersSection(ReadContext &Ctx);
268   Error parseTargetFeaturesSection(ReadContext &Ctx);
269   Error parseRelocSection(StringRef Name, ReadContext &Ctx);
270 
271   wasm::WasmObjectHeader Header;
272   std::vector<WasmSection> Sections;
273   wasm::WasmDylinkInfo DylinkInfo;
274   wasm::WasmProducerInfo ProducerInfo;
275   std::vector<wasm::WasmFeatureEntry> TargetFeatures;
276   std::vector<wasm::WasmSignature> Signatures;
277   std::vector<uint32_t> FunctionTypes;
278   std::vector<wasm::WasmTable> Tables;
279   std::vector<wasm::WasmLimits> Memories;
280   std::vector<wasm::WasmGlobal> Globals;
281   std::vector<wasm::WasmEvent> Events;
282   std::vector<wasm::WasmImport> Imports;
283   std::vector<wasm::WasmExport> Exports;
284   std::vector<wasm::WasmElemSegment> ElemSegments;
285   std::vector<WasmSegment> DataSegments;
286   llvm::Optional<size_t> DataCount;
287   std::vector<wasm::WasmFunction> Functions;
288   std::vector<WasmSymbol> Symbols;
289   std::vector<wasm::WasmDebugName> DebugNames;
290   uint32_t StartFunction = -1;
291   bool HasLinkingSection = false;
292   bool HasDylinkSection = false;
293   bool SeenCodeSection = false;
294   bool HasMemory64 = false;
295   wasm::WasmLinkingData LinkingData;
296   uint32_t NumImportedGlobals = 0;
297   uint32_t NumImportedTables = 0;
298   uint32_t NumImportedFunctions = 0;
299   uint32_t NumImportedEvents = 0;
300   uint32_t CodeSection = 0;
301   uint32_t DataSection = 0;
302   uint32_t EventSection = 0;
303   uint32_t GlobalSection = 0;
304   uint32_t TableSection = 0;
305 };
306 
307 class WasmSectionOrderChecker {
308 public:
309   // We define orders for all core wasm sections and known custom sections.
310   enum : int {
311     // Sentinel, must be zero
312     WASM_SEC_ORDER_NONE = 0,
313 
314     // Core sections
315     WASM_SEC_ORDER_TYPE,
316     WASM_SEC_ORDER_IMPORT,
317     WASM_SEC_ORDER_FUNCTION,
318     WASM_SEC_ORDER_TABLE,
319     WASM_SEC_ORDER_MEMORY,
320     WASM_SEC_ORDER_EVENT,
321     WASM_SEC_ORDER_GLOBAL,
322     WASM_SEC_ORDER_EXPORT,
323     WASM_SEC_ORDER_START,
324     WASM_SEC_ORDER_ELEM,
325     WASM_SEC_ORDER_DATACOUNT,
326     WASM_SEC_ORDER_CODE,
327     WASM_SEC_ORDER_DATA,
328 
329     // Custom sections
330     // "dylink" should be the very first section in the module
331     WASM_SEC_ORDER_DYLINK,
332     // "linking" section requires DATA section in order to validate data symbols
333     WASM_SEC_ORDER_LINKING,
334     // Must come after "linking" section in order to validate reloc indexes.
335     WASM_SEC_ORDER_RELOC,
336     // "name" section must appear after DATA. Comes after "linking" to allow
337     // symbol table to set default function name.
338     WASM_SEC_ORDER_NAME,
339     // "producers" section must appear after "name" section.
340     WASM_SEC_ORDER_PRODUCERS,
341     // "target_features" section must appear after producers section
342     WASM_SEC_ORDER_TARGET_FEATURES,
343 
344     // Must be last
345     WASM_NUM_SEC_ORDERS
346 
347   };
348 
349   // Sections that may or may not be present, but cannot be predecessors
350   static int DisallowedPredecessors[WASM_NUM_SEC_ORDERS][WASM_NUM_SEC_ORDERS];
351 
352   bool isValidSectionOrder(unsigned ID, StringRef CustomSectionName = "");
353 
354 private:
355   bool Seen[WASM_NUM_SEC_ORDERS] = {}; // Sections that have been seen already
356 
357   // Returns -1 for unknown sections.
358   int getSectionOrder(unsigned ID, StringRef CustomSectionName = "");
359 };
360 
361 } // end namespace object
362 
363 inline raw_ostream &operator<<(raw_ostream &OS, const object::WasmSymbol &Sym) {
364   Sym.print(OS);
365   return OS;
366 }
367 
368 } // end namespace llvm
369 
370 #endif // LLVM_OBJECT_WASM_H
371